zj360202的专栏

1.内部表指hive创建并通过load data inpath进数据库的表，这种表可以理解为数据和表结构都保存在一起的数据表。当你通过DROP TABLE table_name 删除元数据中表结构的同时，表中的数据也同样会从hdfs中被删除。CREATE TABLE new_hbase_table(rowkey string, x int, y int) STORED BY 'org.ap

有两种方式写UDF一、就是网上比较

HiveQL 去重操作和SQL一样，HiveQL中同样支持DISTINCT操作，如下示例：(1) SELECT count(DISTINCT uid) FROM log(2) SELECT ip, count(DISTINCT uid) FROM log GROUP BY ip(3) SELECT ip, count(DISTINCT uid, uname) FROMlog G

准备数据SELECT uid, SUM(COUNT) FROM logs GROUP BY uid;hive> SELECT * FROM logs;a 苹果 5a 橙子 3a 苹果 2b 烧鸡 1 hive> SELECT uid, SUM(COUNT) FROM logs GROUP BY uid;

在代码中加载

select * FROM t_xg_product a join t_xg_product_subject_mapping m on (m.product_id=a.product_id)left outer join (select (case when d.product_id is null then '' else cast(d.product_id as string) end

alter table hbase_fct_mcht_item_day add columns(avg_stay_time_bi bigint);会报如下错误：FAILED: Error in metadata: Cannot use ALTER TABLE on a non-native tableFAILED: Execution Error, return code 1 from o

hive> drop table 'fct_pf_vs_hour';FAILED: ParseException line 1:11 mismatched input ''fct_pf_vs_hour'' expecting Identifier near 'table' in table name正常的hql语句执行出问题，解决方法是将 'fct_pf_vs_hour'两边的单引号改成 `f

上面是join的操作 用时进30s，启了两个job，下面是简单操作平均用时15s多些，这里操作数据时70条，不存在数据量的干扰。

1.创建表CREATE TABLE disp_log_data ( ip STRING,time DOUBLE,title STRING,url STRING,hostname STRING,referrer STRING,uvid STRING,isnewvisitor int,scr STRING,spm_pre STRING,spm_url S

报错内容hive> INSERT OVERWRITE TABLE recommend_user_log_detail_show_old_device PARTITION (dt ='2014-08-08' ) > select /*+ mapjoin(recommend_user_score)*/ c2.udid,c2.ad_id,c2.app_id,'0' as log_type,c